Zhemal Khamidun
Head of AI at Alpina Digital, CPO of AlpinaGPT (~30-person engineering & AI team). Founder of Vibe Academy & Hamidun Consulting, author of the AI channel «Готовим ИИшницу» (@jemal_hamidun).
How to Choose an AI Agent for Business: Hugging Face's Open Leaderboard
Hugging Face has released an open benchmark for comparing complete AI agent systems. It found that an agent's architecture matters more than the model behind it.
Cathie Wood Predicts a Robotaxi Boom — the Largest Market for Embodied AI
Investor Cathie Wood says the transportation revolution is the biggest earnings opportunity in embodied AI, and that Tesla and robotaxis are now poised to take off.
Text recognition neural network: PaddleOCR 3.5 adds PyTorch
PaddleOCR now fully supports Hugging Face Transformers as an inference backend, so text recognition and document parsing run in a PyTorch environment.
AI Infrastructure: Cloud GPU Clusters for Business
AI Myths: What Actually Worries an Oxford Professor
How to Control AI Agents: Validating Responses on Bedrock
Controlling AI Spend: How LangSmith Gateway Works
Ettin Reranker: How a Neural Network Improves Search Accuracy
Hugging Face has released 6 Ettin rerankers built on ModernBERT, with state-of-the-art accuracy and speed thanks to Flash Attention 2 and sequence optimization.
Together AI raises Batch Inference API limits 3,000-fold and cuts prices by 50%
The Batch Inference API now handles 30 billion tokens (up from 10 million) and costs half as much as the real-time API. It supports all 40+ models on the platform.
Amazon Bedrock adds three ways to manage AI agent tools
LangChain Deep Agents: How to Choose a Model (ChatGPT, Claude, Google)
How to Train a Large Neural Network for Your Own Tasks: Fine-Tuning DeepSeek
AWS SageMaker: New Tools for Machine Learning
Gemini for Research: An AI Model for Hypotheses and Experiments
Google has unveiled Gemini for Science, a suite of AI tools for researchers. Three features help formulate hypotheses, check experiments and analyze scientific…
How AI agents save memory: LangChain's approach
DeltaChannel, a new primitive in LangGraph 1.2, tackles exponential memory growth in long agent sessions by storing only the changes between steps instead of the…